Application servers executing specialized application software, such as the retail predictive application server (RPAS) from Oracle Corporation, may store permanent records in a binary tree-based multi-dimensional database, which is called a “domain.” The RPAS may be used, for example, to generate forecasts, develop trading plans, and analyze customer behavior. In a customer environment, RPAS users access and store subsets of the domain data in smaller multi-dimensional repositories called “workbooks.” A workbook serves as a sandbox for the user. It allows users to execute business rules, run what-if scenarios, and slice and dice the data to visualize and validate results without requiring continuous access to domain data. Once a user is satisfied with the calculations and results in their workbook, the user commits the data back to the domain. Workbook interactions with the domain include three major operations: workbook build, workbook refresh, and workbook commit. Workbook build refers to the construction of a new workbook and the loading of a subset of the domain data into the workbook. Workbook refresh refers to the refreshing of the data in an existing workbook from the domain. Workbook commit refers to the saving of the workbook data back to the domain.